Predicting output performance of triboelectric nanogenerators using deep learning model

With the development of artificial intelligence (AI), the use of AI algorithms for processing experimental data concerning engineering and physics has attracted broad attention, so the research of triboelectric nanogenerator (TENG) field can also adopt AI technology. Sometimes it is difficult and limited to analyze the impacts of structural parameters on output performance in physical experiments. The reason, on the one hand, is that TENGs structure changes in a small range, which makes the control of experiment difficult; on the other hand, is impossible to complete all the experimental parameters for verifying the experimental rules in the experiment. In this research, an AI algorithm model based upon deep neural networks (DNN) for TENGs is introduced for the first time to predict the output performance of TENGs under various structures and various conditions. The results show that the predicted output power values of the grating, disc, and rolling structures of the TENG by DNN model are consistent with the physical experimental data. Subsequently, we use the DNN model to predict the power output of TENG in the parameter range that the experiment has not tested in the grating, disc and rolling structure. Obviously, this will help researchers analyze the law of the data on a wider parameter range scale, so as to get a better experimental law. At the same time, by adopting the DNN model, we also predict the output performance of the sliding mode TENG under various load conditions.